Generally, linear predictive coding (LPC) is performed on an audio signal having strong speech characteristics. Linear predictive coefficients generated through linear predictive coding are transmitted to a decoder and the decoder reconstructs the audio signal by performing linear predictive synthesis on the coefficients.